Delayed entry into HIV care is a critical problem in the southeastern United States. While advances in antiretroviral therapeutics are dramatically increasing patients' likelihood of achieving virologic suppression and decreasing transmission of virus to others, reduction of HIV in the population is contingent on infected persons first presenting for medical care after diagnosis. Creative strategies to link infected patients with an initial clinic visit are therefore urgently needed. One promising, novel approach to improve linkage to care in HIV is the application of geospatial methods to community-based interventions. Previously, geospatial tools have been used to improve our understanding of HIV transmission dynamics and identification of core areas of transmission, but they have been less used to examine linkage to care. We propose to use modern geostatistical methods to evaluate the clustering of HIV-infected persons who do not link to care (defined as non-presentation to an HIV provider within three months of diagnosis) and to evaluate the relationship of local social contextual factors with non-linkage to care. Specifically, our aims ar 1) To characterize the extent of geographic clustering of HIV-infected persons who do not link to care within the first three months of diagnosis in four large Georgia counties 2) To determine the relationship between community-level socio-environmental factors and non-linkage to care, adjusting for patient-level demographic factors and 3) To describe socio- environmental barriers to care perceived by HIV-infected persons residing in community hot spots of poor linkage to care. To address these aims, we will use data from existing and routinely collected surveillance data for HIV, GA Medicaid Claims, the U.S. Census Bureau, and from publicly available transportation data. Our primary strategy will use spatial cluster analysis to identify areas of poor linkage and multi-level models to determine if place of residence and community-level factors are associated with non-linkage to care. Our models will include geographic parameters such as network distance (based on bus routes and driving times). We will also interview HIV-infected persons residing in areas associated with poor linkage to collect qualitative data on patient perceptions of structural barriers to presentation to care. Our application of spatial methods and multilevel modeling to the linkage to care paradigm for HIV-infected persons will allow a better understanding of the relationship between local geography and linkage to care, and may support a future strategy of geospatially-informed public health interventions.